1 Information-efficient human-computer interfaces David MacKay Department of Physics, University of Cambridge - with David Ward and Alan Blackwell www.inference.phy.cam.ac.uk/mackay/

Slides:



Advertisements
Similar presentations
Data Entry Devices Introduction –Keyboard entry devices are superior to other devices such as knobs, levers, and thumb wheels. –Speed and accuracy are.
Advertisements

Interaction Devices By: Michael Huffman Kristen Spivey.
Copyright 1999 all rights reserved Input Devices n What types are there? n Why do we need them? –What functions do they perform? n What are desirable characteristics.
The Potential of Dwell-Free Eye-Typing for Fast Assistive Gaze Communication PO Kristensson and K Vertanen ACM 2012.
Ahmed Sabbir Arif York University, Toronto, Canada
Lesson 4 Alternative Methods Of Input.
Ambiguous Keyboards 4/8/ Nate Bodenstab Text Entry for Mobile Computing: Models and Methods, Theory and Practice. MacKenzie and Soukoreff. Human-Computer.
Dialogue Design Speech, pen, and gestures Speech Output  Tradeoffs in speed, naturalness and understandability  Male or female voice? Technical issues.
Efficient Computer Interfaces Using Continuous Gestures, Language Models, and Speech Keith Vertanen Inference Group August 4th, 2004.
Text Input: Techniques and Research Tools Poika Isokoski at NIT Background: A Collage of images scanned from: Albertine Gaur. A history of.
1 Dasher - A Fast and Easy Keyboard Alternative David MacKay Department of Physics, University of Cambridge with David Ward, Iain Murray, Matthew Garrett,
1 Lab Session-IV CSIT-120 Spring 2001 Lab 3 Revision and Exercises Rev: Precedence Rules Lab Exercise 4-A Machine Language Programming The “Micro” Machine.
Chapter 8_2 Bits and the "Why" of Bytes: Representing Information Digitally.
Using CTW as a language modeler in Dasher Phil Cowans, Martijn van Veen Inference Group Department of Physics University of Cambridge.
People & Devices: (Inputs & Outputs) Startlingly small child using computer History of human-computer interaction Another history video.
TAUCHI – Tampere Unit for Computer-Human Interaction 1 Alternate Keyboards for Text Entry – And How to Evaluate Them I. Scott MacKenzie.
Topic: Speaking General objectives: Students will be able to teach speaking. Students will be able to integrate speaking with listening, reading and writing.
1 Lab Session-III CSIT-120 Spring 2001 Revising Previous session Data input and output While loop Exercise Limits and Bounds GOTO SLIDE 13 Lab session.
Using CTW as a language modeler in Dasher Martijn van Veen Signal Processing Group Department of Electrical Engineering Eindhoven University.
Lesson Objectives To understand that users with disabilities require different input and output devices To be able to identify these devices and explain.
Chapter 3 Computer Science and the Foundation of Knowledge Model
 A data processing system is a combination of machines and people that for a set of inputs produces a defined set of outputs. The inputs and outputs.
Text Input to Handheld Devices for People with Physical Disabilities Brad A. Myers and Jacob O. Wobbrock Human Computer Interaction Institute School of.
A Virtual Keyboard with Multi Modal Access for people with disabilities Vijit Prabhu 1, Girijesh Prasad 2 1 Computer Science & Engineering, Indian School.
CSCI 1200 / ASSC 1000 Software program : instructions for hardware. makes computers useful. early computers : re-wiring. today : programming.
1 Lab Session-III CSIT-120 Fall 2000 Revising Previous session Data input and output While loop Exercise Limits and Bounds Session III-B (starts on slide.
TiltText: Using Tilt for Text Input to Mobile Phones Daniel Wigdor & Ravin Balakrishnan.
Interaction Gavin Sim HCI Lecture /111. Aims of this lecture Last week focused on persona and scenario creation. This weeks aims are: ◦ To introduce.
Handhelds in the Automobile: The Denali Navigation System Human Computer Interaction Institute School of Computer Science Carnegie Mellon University Brad.
Dialog Design - Gesture & Pen Interfaces, Mobile Devices CS / Psych This material has been developed by Georgia Tech HCI faculty, and continues.
Gaze-Controlled Human-Computer Interfaces Marc Pomplun Department of Computer Science University of Massachusetts at Boston Homepage:
22CS 338: Graphical User Interfaces. Dario Salvucci, Drexel University. Lecture 10: Advanced Input.
Input Devices Keyboards (Notes in Text). The keyboard is an important peripheral that is used as an input device of a computer and many other devices,
KAMI KITT ASSISTIVE TECHNOLOGY Chapter 7 Human/ Assistive Technology Interface.
Recent Results in Combined Coding for Word-Based PPM Radu Rădescu George Liculescu Polytechnic University of Bucharest Faculty of Electronics, Telecommunications.
1 Human Computer Interaction Week 5 Interaction Devices and Input-Output.
4 November 2000Bridging the Gap Workshop 1 Control of avatar gestures Francesca Barrientos Computer Science Division UC Berkeley.
Question 14 Exercise page 341 Carwash. This records our frustration with trying to match our answer with the back of the book. Learning did happen.
Representing Characters in a Computer System Representation of Data in Computer Systems.
Initial Product Scope Developing textual input methods for new devices. Michael Dann.
HMM finds behavioral patterns… Zoltán Szabó Eötvös Loránd University.
Pen Based User Interface Issues CSE 490RA January 25, 2005.
SONGONUGA EMILIA ACCOUNTING 12/SMS02/ Introduction One goal of human-computer interaction research is to reduce the demands on users when using.
Human Computer Interaction Lecture 4 The Computer.
UWave: Accelerometer-based personalized gesture recognition and its applications Tae-min Hwang.
Information theory Data compression perspective Pasi Fränti
Nat 4/5 Computing Science Data Representation Lesson 3: Storing Text
Lesson 4 Alternative Methods Of Input.
3.3 Fundamentals of data representation
Alternative Methods Of Input
Standard Methods of Input.
Jacob R. Lorch Microsoft Research
Dialog Design 3 How to use a PDA
Input Devices Text Entry Devices
Franklin (Mingzhe) Li, Mingming Fan & Khai N. Truong
Lesson 4 Alternative Methods Of Input.
Condensed Lesson Plan*
F-Pointer: Prototype testing of the finger-manipulated device
Planning with PDAs Copyright 2006 South-Western/Thomson Learning.
NBKeyboard: An Arm-based Word-gesture keyboard
The Morse Mouse Crossing Interface
Introduction to Computers
Data Representation Conversion 05/12/2018.
Source Encoding and Compression
Multimodal Human-Computer Interaction New Interaction Techniques 22. 1
Lesson 4 Alternative Methods Of Input.
Learning Intention I will learn how computers store text.
Introduction to Computers
Human and Computer Interaction (H.C.I.) &Communication Skills
KEYBOARDING: SPEED & ACCURACY
Presentation transcript:

1 Information-efficient human-computer interfaces David MacKay Department of Physics, University of Cambridge - with David Ward and Alan Blackwell

2 Information-efficient text entry David MacKay Department of Physics University of Cambridge - with David Ward and Alan Blackwell

3 Hands-free writing Information-efficient human-computer interfaces David MacKay Department of Physics - with David Ward and Alan Blackwell

4 Dasher – A Data Entry Device Using Continuous Gestures and Language Models David Ward, Alan Blackwell, and David MacKay University of Cambridge Original concept: MacKay and Lewicki, 1997

5 A famously inefficient writing method Alternative keyboard layouts

6 Text entry for handheld devices Miniature or rearranged keyboards Gestural alphabets Unistrokes

7 Dynamic selection Word-completion Write ambiguously, disambiguate later  T9  Spellchecker  Shorthand dsmbgu8 l8r Quikwriting

8 What’s wrong with keyboards?

9 1. Information content of English  1 bit per character. Each keypress on a QWERTY keyboard could convey ~ 6 bits. 2.Keyboard – digital : hands - analog. A pointing finger can generate information at a rate of 14 bits per second (Drury and Hoffmann). So... Potential writing speed of just one finger is 14 characters per second? (  170 words per minute) Why keyboards are inefficient

10 Writing and text-compression Text compression Text Bit string (preferably short) Gesture (preferably brief) Text Writing

11 Writing and text-compression Optimal text compression – Arithmetic coding Text Bit string, viewed as a real number Text Real gesture Writing with Dasher probabilistic model

12 Demonstration - available for GNU/linux, windoze, and pocket PC

13 Arithmetic Coding a c P(x 1 =a) P(x 1 =c) P(x 1 =b) a b c P(x 1,x 2 )=P(x 1 )P(x 2 |x 1 ) P(x 1,x 2,x 3 )=P(x 1 )P(x 2 |x 1 )P(x 3 |x 1,x 2 ) P(x 1 =a,x 2 =a) P(x 1 =a,x 2 =c) String S=x 1 x 2 x 3... Divide the interval (0,1] into intervals equal to the probabilities of the symbols. 0 1

14 Dynamics Point to where you want to go  Like driving a car  Motion sickness? Passengers may get sick, driver doesn’t

15 Benefits Keyboard – usually one gesture per character Dasher – some gestures select more than one character Inaccurate gestures can be compensated for by later gestures

16 Benefits continued Mode-free. Can be used with any alphabet (e.g. Hiragana!) Requires no special learning. (knowledge of the chosen alphabetical order is helpful) Can add extra characters to alphabet without any extra learning.

17 The Language Model Based on PPM (Prediction by Partial Match), a context-based model. Compresses most English to about 2 bits per character (could be improved) Fast Adaptive Works with any language

18 Evaluation 10 volunteers Dictation task Emma, by Jane Austen Automated dictation system with recorded speech 12 Dasher exercises, each 5 minutes long Keyboard exercises between Dasher sessions Measured writing speed and word error rate Dasher Keyboard 5 min3 min

19 Results - writing speeds Writing speed (cpm) Exercise number Writing speed (cpm) Exercise number DasherKeyboard 50 wpm 25 wpm

20 Results – information rate

21 Results – writing errors Percentage of words wrong Exercise number Percentage of words wrong DasherKeyboard

22 Comparison with other devices Dasher Bellman OPTI Half-QWERTY ABC-tapping Hand printing QWERTY-tapping TCK2 Chorded, one hand one hand keyboard stylus on large tablet Stylus fluctuating keyboard

23 Mobile Text Entry Pocket PC Driven by stylus & touchscreen

24 Palmtop Results

25 The main defect of Dasher It demands visual attention (like any predictive system)

26 Eye gaze tracking

27 Hands-free writing

28

29 Eyetracking results

30 Eyetracking results

31 Another hands-free solution Head-mouse

32 Different Character Sets Include capital letters, punctuation Japanese – Hiragana

33 Exploits redundancy of language Rapid learning rate Applications - Mobile text entry - Special needs Download Dasher! - available for linux, windoze, and iPaq handheld Summary of Dasher

34

35 Download Dasher - available for linux, windoze, and pocket PC Development of Dasher is supported by the Gatsby Charitable Foundation

36 Some Uses for Probabilities “Bayesian Methods for Adaptive Models” Fast approximate inference Error-correcting codes Efficient data-entry devices